utf8 isn't flakey, j is you can still get your numbers back if you try hard enough
]t=: 3 u: 8 u: 97 243 98 97 195 179 98 3 u: 9 u: 1 u: 10 u: t 97 243 98
Converting utf8 (literal) to unicode or unicode4 prior to counting/appending them with other unicode should stay sane(?)
Yes, but the question is: why is there a format which isn't sane, and why do I have to explicitly select the sane format?
---------------------------------------------------------------------- For information about J forums see http://www.jsoftware.com/forums.htm
